# Robyn Dawes The psychologist who **demolished the case for expert clinical judgment with simple linear models**. Dawes (1936–2010), longtime Carnegie Mellon professor, conducted some of the most empirically devastating research in social science: simple weighted-sum statistical models routinely outperform expert clinical judgment on prediction tasks across medicine, parole, admissions, and personnel selection. The findings have been replicated across decades and domains and are largely ignored by the institutions whose practices they implicate. ## Core Contributions ### Clinical vs. Actuarial Prediction The core finding, demonstrated across many studies: when predicting outcomes (medical diagnosis, college performance, parole recidivism, marital stability, hiring success), simple linear formulas combining a few measurable inputs reliably outperform the holistic judgment of credentialed experts. The expert sees more variables, more context, and more nuance — and is *less* accurate. The actuarial formula sees fewer variables but combines them consistently, while the expert combines them inconsistently and is influenced by irrelevant factors (recent news, mood, similarity to remembered cases). ### "Improper" Linear Models Dawes's most striking finding (1979): even *poorly weighted* linear models — including models with random or equal weights — outperform expert judgment on most prediction tasks. This is not because the weights are correct but because *consistency* matters more than *optimization*. Experts are inconsistent; even bad models are consistent; consistency wins. ### House of Cards (1994) Dawes's book *House of Cards: Psychology and Psychotherapy Built on Myth* extended the clinical-vs-actuarial findings into a sustained critique of the psychotherapy profession. The argument: licensure, credentialing, and clinical experience produce essentially zero gain in therapeutic outcomes; what works (when anything works) is technique, not the therapist's judgment. The book made Dawes unpopular within his profession but the empirical record has held up. ## KB Significance **1. Expertise vs. accuracy as a recurring asymmetry.** Dawes's findings inform a broader pattern visible across the RWKB: credentialed expert judgment is often *less* accurate than simple structured procedures, especially in domains where the expert's reputation depends on appearing decisive. Pundit prediction, intelligence-community estimates, and professional-class confidence all show the pattern Dawes documented. The audience-trust calibration is structurally wrong. **2. The actuarial frame for political prediction.** [[Philip Tetlock]]'s work on superforecasting is the contemporary continuation of Dawes's program: expert political judgment is on average barely better than chance, and structured procedures (consistent decomposition, calibration tracking, base-rate anchoring) outperform unstructured expert intuition. Dawes is the deeper foundation; Tetlock the empirical extension into the political domain. **3. The case against pure judgment-trust.** A specific RWKB-relevant implication: institutions that resist documentation, structured procedures, and outcome-tracking — preferring to rely on the judgment of credentialed personnel — are organizationally prone to the failures Dawes documented. The right-wing project's preference for personal authority over institutional procedure is, among other things, a preference for the less-accurate decision-making mode. [[Schedule F]]'s destruction of civil-service procedure in favor of presidential discretion is the political form of this preference. ## See Also - [[Daniel Kahneman]] — Closely related research program; Kahneman's *[[Thinking Fast and Slow]]* discusses Dawes extensively - [[Philip Tetlock]] — The political-prediction extension of Dawes's findings - [[Falsifiability]] — The discipline of submitting judgment to outcome-feedback that Dawes's framework operationalizes - [[Decision Science]] — The hub where this work lives